The major premise of this application is that an adequate ability to learn and to remember is fundamental to normal cognitive behavior and the minor premise is that many cognitive behaviors hinge upon declarative memories that require a hippocampus for appropriate storage. From these premises follow the long term goal of understanding how the hippocampus initially forms declarative memories and then interacts with cerebral cortex in the storage of long-term memories. The more immediate goal is to provide a quantitative understanding of hippocampal function by simulating biologically plausible hippocampal-like networks in cognitive- behavioral situations that require the hippocampus for their normal function. Such simulations, in a particularly surprising way, show the same sensitivity to training procedures as do rats and humans. Namely the model predicts the distribution of individual learned performances. Such models might therefore be used to develop optimal learning/training procedures to improve the poorer learners. Using three paradigmatic learning problems and a spectrum of closely related, minimal, biologically plausible models of the hippocampus, the specific aims of this proposal are: 1) to understand information processing in the hippocampus including its critical biological substrates; 2) for each learning paradigm, to predict the patterns of hippocampal cell firing that occur during learning, during rest periods over the course of learning, and during testing after learning; 3) based on the individual differences that arise from the parameterized biology of the model, to explain the individual differences of behavioral performance in animals; and 4) to prove (or improve, if unsuccessful) the viability of the model by predicting behavioral outcomes in novel training situations. Computer simulations will be performed of trace conditioning and of two cognitive paradigms--transitive inference and transverse patterning. Using mathematical analyses and simplified models of the hippocampus, which are systematically varied in both complexity and parameterization, the applicant will attempt to understand how the archetypal hippocampal anatomy and associated physiologies reproduce the functions ascribed to the hippocampus [including context formation and flexible memory representations (Eichenbaum et al., '92)] and why some other biologies and parameterizations fail.